Discriminating Power of an Sample Entropy and a Nonlinear Association Index in Prediction of a Preterm Labor Based on Electrohysterographical Signals

Recognition of physiological and pathological patterns in biomedical signals is still a challenge. The are many linear and nonlinear techniques proposed for this purpose but their effectiveness is seldom compared and ranked. The aim of the paper was to compare a discriminating power of an sample entropy and a nonlinear association index in prediction of a labor based on electrohysterographical (EHG) signals. The EHG signals were registered at women being during a labor or waiting for beginning of a labor. A sample entropy was estimated for a single component of an EHG signal. A nonlinear association index was computed to express a plausible relation between two components of an EHG signal. The comparison of usefullness of these parameters in a labor prediction was performed using ROC. The obtained results show that a labor prediction based on the nonlinear association index is more effective than using the sample entropy.